Résumé
1. Introduction Sub-Saharan Africa (SSA) faces significant food availability issues largely due to low soil fertility leading to low crop yields. Climate change is expected to exacerbate these issues due to a combined negative impact on crop yield and soil fertility. Integrated soil fertility management (ISFM) implies the combined use of mineral and organic fertilizers and is seen as a solution to increase soil fertility and crop yield, yet it is uncertain how this advantage is maintained with climate change. Soil-crop models are useful tools to assess the performance of ISFM. Multi-model simulations are often more robust than single-model simulations and contribute to understanding and reducing modelling uncertainty (Falconnier et al., 2020). In this study, we aim to i) evaluate multi-model performances for long-term simulations of crop yield and soil organic carbon (SOC) and their feedbacks (Couëdel et al., 2024), ii) use the calibrated multi-model ensemble to explored the impact on maize yield for scenarios of change in soil fertility and climate variables in four representative sites in SSA with no input and ISFM management (Couëdel et al., submitted). 2. Materials and Methods We compared the performance of 16 soil-crop models using data from four long-term experiments at sites in SSA with contrasting climates and soils. Each site had experimental treatments including no exogenous inputs and ISFM. Once calibrated and evaluated we used the multi-model ensemble under reset and continuous simulations over 30 years to assess the impact of soil fertility vs change in climate variables (temperature, rainfall, and CO2) on crop yield. In reset simulations, SOC, soil nitrogen and soil water were reinitialized each year with the same initial conditions. In continuous simulations, SOC, soil N and soil water values of a given year were obtained from the previous year's simulation, allowing for cumulative effects on SOC and crop yields. 3. Results Model ensemble evaluation show that uncertainty increased over the duration of the long-term experiments. SOC simulations uncertainty was largest when organic amendments were applied, whilst yield predictions uncertainty was largest when no inputs were applied. We found discrepancies among models in simulating soil-crop feedbacks due to uncertainties in simulated carbon coming from roots and simulated crop N supply from soil organic matter decomposition. Yet when used as an ensemble, the 16 models enable to reproduce with satisfactory accuracy yield and SOC dynamics in all sites. In the scenario analysis, most models agreed that with baseline management (no input) the magnitude of yield changed was much larger when considering declining soil fertility with baseline climate (-39%), compared with considering constant soil fertility but changes in temperature, rainfall and CO2 (from -12% to +5% depending on the climate variable considered) (Figure 1.A). The model ensemble showed that when changes in soil fertility were taken into account, the advantages of ISFM systems over no-input systems increased over time (+190%) (Figure 1.B). This increase in ISFM benefits was greater at sites with low initial soil fertility. 4. Discussion We found that soil fertility changes had more impact than climate change on maize crop yield in SSA. We argue for the urgent need to consider long-term soil-crop feedbacks in climate change studies to avoid largely underestimating the impact of climate change and ISFM on food production in SSA. The model evaluation also emphasizes the need for long-term experiments in which root and soil N dynamics are monitored. This will provide the relevant data to improve and calibrate soil-crop models, leading to more robust and reliable simulations of SOC and crop productivity, and their interactions.